TrigMC software release coordination: PTD's logbook - page for July 2018

Notes

Use reverse chronological order.

For each shift date show the NEW MRs, and any related relevant comments/questions; note down when I (or some one else) merged/closed each request, and mark the MR as done/closed by ticking the box ([x]). If an MR takes longer than a day to merge, leave it listed under the original shift date, but add comments prefaced with the date of the comment/action. Only MRs that are still active should have the box unticked ([ ]).

Therefore each MR will only appear once in the log, under the date of the day when I first became aware of it.

30/7/2018

[x] Release 21.5.7,Athena off 21.0-TrigMC,Athena,r2018-07-29T2139 (requested by John Baines)

10.07am John Baines has confirmed that he would like a release made out of 21.0-TrigMC,2018-07-29T2139. (Will start working on it as soon as my meeting with Adam - now - is over.)

11.40am Contact Oleg Kuprash, just to check he sees no impediment to making the above release. Started the Jenkins job to copy the nightly build files over to the CVMFS location, for installation.

12.13pm Oleg has given green light for the release. Will crete JIRA ticket now.

12.20pm JIRA ticket opened (ATLINFR-2549), assigned to Alessandro de Salvo.

12.59pm Have now also completed: tagging the new release, and updating the release candidate number (it turns out that I do not, as previously suspected, have push privileges into 21.0-TrigMC, so the only way to accomplish this is via an MR, which I merged immediately and also cancelled the running of the pipeline).

31/7/2018 Alessandro took longer than usual to respond (it turns out that he, too, is on holiday), but after prompting he got the installation going. Informed John.

29/7/2018

[x] !13105 Update 21.0-TrigMC to 21.0.76 (opened by Oleg Kuprash)

2.18pm MR opened by Oleg 4h ago. Pipeline running.

2.19pm Pipeline has just now (!) concluded OK. Merged OK.

28/7/2018

1.30pm TO NOTE: MR !13037 (https://gitlab.cern.ch/atlas/athena/merge_requests/13037) is labelled "alsoTargeting:21.0-TrigMC". This was at the specific request of John Baines, who wants to have this included for a TrigMC production. This MR was successfully merged about 20h ago. Therefore keep an eye on when it is actually swept into TrigMC (normally happens in the following 24h period), and then ask John if he wants a release made out of it.

1.35pm Actually, have just realised (reading the entire MR page carefully) that Atlas Nightlybuild has posted that the MR was swept successfully into 21.0-TrigMC (and into 21.0) 11h ago (but the sweep to "master" failed). Ask John how he wants to proceed.

1.50pm Slightly confused: even though 13037 specifically states "Sweep successful ... into 21.0-TrigMC" it seems (from the timings) that this is exactly when the related MR (actually carrying out the sweep, 13092, see below) was opened. (And that MR has failed... so it seems that the sweep was not successful at all...). So it could be that the "Sweep successful" message, simply indicates that it was successful in creating the MR to do the sweep (?).

[x] !13092 Sweeping 13037 from 21.1 to 21.0-TrigMC. Development for ATR-18398 (opened by Atlas Nightlybuild)

12.59 Another MR opened last night, at the same time as the one below. Pipeline has failed "make" but passed all other steps... Will look into it.

1.24pm The make fail is related to a symbolic link issue (trying to make a link when that link (or a file with the same name: chainDump.py) already exists in that directory... Not obvious to me that this isn't a coding error that needs fixing (but equally it could be that , fro some reason, the directory where this happens wasn't properly cleaned before the make started...). As it is the weekend, and not much might happen, it is probably worth rertrying for a build, in case that will actually work. therefore asked for a another build. Pipeline started within a minute.

9.23pm Pipeline completed successfully. Merged OK.

[x] !13089 Sweeping 13027 from 21.0 to 21.0-TrigMC. small tweak to SSB hardware constants generation (opened by Atlas Nightlybuild)

12.54pm This MR was opened last night, around 2am. Pipeline completed successfully. Merged OK.

18/7/2018

[x] !12806 Sweeping !12397 from 21.0 to 21.0-TrigMC. Update FTK AOD creation for data BS files (opened by Atlas Nightlybuild)

8.04am Request appeared overnight. Pipeline has already completed successfully. Merged OK.

17/7/2018

[x] New TrigMC release out of r2018-07-09-T2159 (requested by John Baines)

6.35pm Around lunchtime today John requested that a new release be made out of a specific nightly (more than one week old: 9/7/2018). It is the first time that I do this for 21.0-TrigMC, under the new Gitlab system. Oleg provided a link to the instructions on how to do this. I spent some time working through the instructions, and then completed the three steps specified there: (i) Create a Jira ticket for CVMFS installation; (ii) Tag the release in git; (iii) Update the release candidate number. The actual JIRA ticket requesting the release went in at 5.48pm, and was assigned to Alessandro de Salvo; I have not heard anything back yet. I have emailed John and Oleg 10 minutes ago to let them know that the request has been submitted.

18/7/2018 8.19am On JIRA: Oana posted a message last night to tell Alessandro that the RPMs have now been copied to the official location, so that Alessandro can deploy the release. She also mentioned how I can do this the next time (see my JIRA ticket). No reply from Alessandro yet.

18/7/2018 2.34pm Alessandro has now confirmed that the installation is complete and has closed the JIRA ticket. I have emailed John and Oleg to let them know as much.

[x] !12815 version.txt: Update release candidate number to 21.5.7 (opened by myself)

6.41pm This MR was triggered by step (iii) above. CI pipeline has started (but is it really needed to waste resources on a pipeline for this entirely trivial change? Should I simply "Merge Immediately"?)

6.58pm Have looked in git for past MRs of this type (i.e. simply updating version.txt). In some cases the pipeline has been allowed to run and only then was the MR approved, but also found several cases where it seems that the MR was submitted and immediately merged, which seems to have avoided the pipeline starting. Therefore decided to approve this MR now -- seems to have merged OK. (Nevertheless, the pipeline continues to run... Next time, submit it and then immediately approve, to see if that avoids the pipeline starting in the first place.)

18/7/2018 8.10am Just out of curiosity, checked the pipeline status: completed successfully.

15/7/2018

[x] !12721 sweeping MR 12251 to 21.0-TrigMC (opened by Tulay Cuhadar Donszelmann)

1.03pm Request was opened Friday, late. Only saw it now (Sunday). Simple sweep. Will merge next.

1.04pm Merged OK.

10/7/2018

[x] EventOverlayJobTransforms-00-06-32-13 for the next rel. 20.7.9.8.9, TrigMC (Requested by Andrzej Olszewski)

6.30pm (CERN) Andrzej asks that the above tag be added to TrigMC 20.7 cache, and that a build be requested immediately, as he has done all required validation and testing.

6.47pm Branch tag added to the cache 20.7.9.8.8 (not 20.7.9.8.9, as AO wrote in his request) and marked as "accepted". Will ask for immediate build now.

7.21pm Update: it turns out that the next cache should indeed be 20.7.9.8.9... 20.7.9.8.8 was built and released almost a month ago, but - for some reason - the cache was not closed and the new one (20.7.9.8.9 opened), as requested... Therefore, I have had to revert 20.7.9.8.8. to the earlier tag (EOJT...-32-11, as it turns out), and have now asked the Release Shifters again to close this cache, open 20.7.9.8.9, and sweep tags from 20.7.9.8.8 to this latest one. Only once this is done can I add the EOJT-...32-13 tag to the cache and then proceed with requesting a build and release... Waiting.

7.39pm Oana Vickey Boeriu has responded that the above requests are now done. Will check and proceed with request for new release asap.

7.45pm I have now added the EOJT-...32-13 tag to the TrigMC 20.7.9.8.9 cache and marked it as accepted.

7.50pm I have now emailed the release shifters asking to make a release out of the 20.7.9.8.9 cache, then close it etc...

7.54pm e-mailed Andrzej Olszewski et al to let them know we are now back on track.

11/7/2018 (CERN) 8.55am In email contact with Oana, to confirm that the release(s) from last night showing in NICOS are OK, and that she can proceed with distribution, etc. Note that there are two rel_3 showing: I am guessing that one was triggered by Oana making a release out of TrigMC 20.7.9.8.9 and the other is the habitual nightly (they should therefore be identical - but at the moment one of them is still not finished, in the process of completing the tests).

11/7/2018 11.48am Oana has confirmed the release is now available and is awaiting Grid distribution. Have emailed the developers and atlas-trig-rel-coord accordingly.

11/7/2018 11.52am Also: have checked that 20.7.9.8.9 has been closed and 20.7.9.8.10 opened, as requested.

3/7/2018

[x] !12484 Update xAODBTaggingEfficiency file to recommended for r21. Fix for ATLASRECTS-4511 (opened by Tim Martin)

12.26pm CI pipeline failed (...! Tim M describes it as a "bit of a chicken-and-egg situation"; see !12357 in June 2018 logbook page...)

1.31pm Julie requested that I go ahead and Merge, without waiting for CI pipeline, to get out of this catch-22 situation. Merged OK.

-- PedroTeixeiraDias - 28 Jun 2018